Objectives: The aim of this pilot study was to demonstrate the potential of simultaneously acquired 68-Gallium-DOTA-D-Phe1-Tyr3-octreotide (68Ga-DOTATOC) positron emission tomography/magnetic resonance imaging (PET/MRI) in comparison with 68Ga-DOTATOC PET/computed tomography (PET/CT) in patients with known gastroenteropancreatic neuroendocrine tumors (NETs).
Materials and methods: Eight patients (4 women and 4 men; mean [SD] age, 54  years; median, 55 years; range 25-74 years) with histopathologically confirmed NET and scheduled 68Ga-DOTATOC PET/CT were prospectively enrolled for an additional integrated PET/MRI scan. Positron emission tomography/computed tomography was performed using a triple-phase contrast-enhanced full-dose protocol. Positron emission tomography/magnetic resonance imaging encompassed a diagnostic, contrast-enhanced whole-body MRI protocol. Two readers separately analyzed the PET/CT and PET/MRI data sets including their subscans in random order regarding lesion localization, count, and characterization on a 4-point ordinal scale (0, not visible; 1, benign; 2, indeterminate; and 3, malignant). In addition, each lesion was rated in consensus on a binary scale (allowing for benign/malignant only). Clinical imaging, existing prior examinations, and histopathology (if available) served as the standard of reference. In PET-positive lesions, the standardized uptake value (SUV max) was measured in consensus. A descriptive, case-oriented data analysis was performed, including determination of frequencies and percentages in detection of malignant, benign, and indeterminate lesions in connection to their localization. In addition, percentages in detection by a singular modality (such as PET, CT, or MRI) were calculated. Interobserver variability was calculated (Cohen's κ). The SUVs in the lesions in PET/CT and PET/MRI were measured, and the correlation coefficient (Pearson, 2-tailed) was calculated.
Results: According to the reference standard, 5 of the 8 patients had malignant NET lesions at the time of the examination. A total of 4 patients were correctly identified by PET/CT, with the PET and CT component correctly identifying 3 patients each. All 5 patients positive for NET disease were correctly identified by PET/MRI, with the MRI subscan identifying all 5 patients and the PET subscan identifying 3 patients. All lesions considered as malignant in PET/CT were equally depicted in and considered using PET/MRI. One liver lesion rated as "indetermined" in PET/CT was identified as metastasis in PET/MRI because of a diffusion restriction in diffusion-weighted imaging. Of the 4 lung lesions characterized in PET/CT, only 1 was depicted in PET/MRI. Of the 3 lymph nodes depicted in PET/CT, only 1 was characterized in PET/MRI. Interobserver reliability was equally very good in PET/CT (κ = 0.916) and PET/MRI (κ = 1.0). The SUV max measured in PET/CT and in PET/MRI showed a strong correlation (Pearson correlation coefficient, 0.996).
Conclusions: This pilot study demonstrates the potential of 68Ga-DOTATOC PET/MRI in patients with gastroenteropancreatic NET, with special advantages in the characterization of abdominal lesions yet certain weaknesses inherent to MRI, such as lung metastases and hypersclerotic bone lesions.